From my point of view, if Storj changed our price to this number, it will be the last day with Storj and I personally will say: “Good-by my dear friend! time to die”.
So, I think nothing to discuss here, will see the real actions from Stoj-Lab. But I can promise if our price is dropped, a lot of storage nodes will down because much more profitable to do nothing than support nodes with this price.
12 Likes
I’ll just say.
My 0.5PB will be transferred offline.
I won’t even cover the cost.
It will be cheaper to just turn everything off.
All my expenses over the past few years have only increased. Nothing gets cheaper.
3 Likes
I completely agree with you. My tens of terabytes will also be instantly turned off without a graceful exit. In general, prices should not be reduced, but increased due to global inflation. In my opinion, prices should be at least 20% higher than current levels.
4 Likes
It’s not sustainable. Increase prices for customers → no more customers → no traffic.
However, since at least my costs are mostly independent of traffic (idle server uses about as much power as uploading at 100mbps), if the rate went down, but egress went up to compensate it would be good.
If payment for egress went down from $20/TB to $5TB, but egress went up from 5.8mbps (my average for the past month) to 25mbps, it would even be OK.
Similar with stored data - my server has 12 free slots for hard drives, I can expand the pool.
1 Like
with expanding your setups you will expand electricity costs also. in my setup with 67 hdd, it is big already
Now I have a question what Storj is effectively paying to SNOs and charging their customers.
When I store a 1TB file on Storj DCS:
- Storj does not charge for upload bandwidth, Storj does not pay for upload bandwidth.
- With expansion factor of 2,7, what does Storj charge for storage? 1 TB or 2,7 TB? Surely they have to pay for 2,7 TB to SNOs for a 1 TB file.
- Now the egress. With 29/80 reconstruction ratio is it true that I will have to download effectively only 362,5 GB of my entire 1 TB file to get the full file? What does Storj charge? 1 TB file size bandwidth or 362,5 GB real bandwidth? Surely they would only have to pay for 362,5 GB to SNOs if this is what would be the real used bandwidth.
Customers pay for 1 TB. The expansion factor is a property of the network. Customers shouldn’t have to pay for how Storj ensures redundancy. So yeah, they charge customers for 1TB and pay node operators for 2.76TB.
No, you’re applying the expansion factor twice there. You have to download any 29 pieces which would total the original 1TB. The 80 pieces would be 2.76TB total.
4 Likes
Very interesting discussion. I want to add my feedback, as every SNO should, because it helps everybody.
My setup:
I’m operating a few nodes in different locations, my oldest hitting 2 years soon. None of them are full. I’m planing to alocate 180+ TB for Storj, untill hardware dies, and maybe continuing with new one if it’s worth it. Raspberryes are to complicated for me, PC/Servers are to huge for the majority of locations and draw to much power, etc. I’m only using brand new hardware (2 bay NAS, Exos HDD, UPS) and fiber optic internet at every location. The weakest link is 300/150Mbps, and the strongest 900/450Mbps.
I’m not interested in pros and cons of using brand new/expansive hardware vs anything else, because… that’s my choise and I could afford it.
My expenses for 6 years, per location, are… maybe 
: NAS + 2 HDDs (2x16TB) + UPS + 1 battery replacement (1 per 3 years) + energy (hard to tell… 3$/NAS/month for now, not sure what it will be with 2 HDDs runnig, just installed 1 in each NAS, for now, and the energy prices are changing from month to month). I din’t mentioned routers and internet, because they are already installed and used for usual activity. 1 complete setup is aprox. 1250$ for 28 TB alocated space. My oldest node started in ian. 2021, with only 5 days of downtime, just reached 30$/month, and has made 230$, including oct. 2022. So the ROI is very promising. I deserve the Nobel in economics. 
Feedback
Laughing aside, this is the most professional I can be as a SNO. I don’t know what Storj team has in mind when talks about professional certified hardware, but this is the best I can bring to the table. Dedicated servers are out of the question… or anything else.
For october, the total payout is 70% from engress-download, 30% from repair (insignificant) and storage occupied. So cutting down payment for engress will be a huge blow for everyone with not full nodes. When the node is full, the payouts from engress-download are going down quickly.
My suggestions would be:
- FIRST THING! STOP selling us short! Storj network is already a reliable cloud storage network, and the most secure in the world. None of others have the distribution, encryption and redundancy of Storj. The continuous developement, the dev superteam, the tens of github commits per month says “Man, this service is well mantained. I can trust my data to them.” You should ask bigger prices for storage and engress. We are not in infancy anymore.
- more marketing! So many potantial clients are waiting there, but don’t know about the benefits of Storj DCS and the differences from the competition. A huge nice poster with the exact differences, advantages and disadvantages from the others, nominating exactly who is there and what it offers, and what Storj offers, will be huge marketing for us. Easy to read and understand, with all the necessary specs, for newbies and experts. Maybe on the main page of Storj site, on Tweeter, etc etc.
More YT videos, social media adds, etc, more marketing… - a new “held back/collateral” scheme for SNO; every month, an 1-5% percent of revenue will be held as collateral. This will be used to pay the repairs, if youre node goes down for good. If you gracefully exit, only half of this collateral will be returned. If no gracefull exit, no collateral returned. This will stimulate SNOs to use better and reliable hardware. The held amount is insignifiant right now.
- implement tiers for engress-download SNO payouts; ex.: for 0-500GB - 30$; for 500-1000GB - 20$; over 1TB - 10$.
- stop sending test data and maybe delete some; paying for not used space, “just in case”, maybe is not the best for Storj right now.
- implement an easy way to opt-out or opt-in for receiving test data, like a simple parameter “-- test-data ON/OFF”. White-listing some sats and black-listing others maybe is not a smart move. In the future, the test satellites may become client-data satellites, and if you miss out the announcement, you will miss out client-data. The parameter should only block the test data, untill the test sats become client-data sats.
- I disagree with “data center” like professional SNOs, because they will receive more traffic per /24 subnet than the usual SNO, to make it worth their resources, and it will reduce the trafic for us. Offer the same rules to everyone. Encurage use of better and reliable hardware instead.
- an efficient “proposal distribution system”, dubbled by an easy-to-use voting and feedback system, that informs every SNO of the future developements, with transparency/ no secrecy, and gives him the possibility of tell his opinion in short. We all put our emails in when we create nodes. Why don’t use them? There are not many SNOs reading the forum daily, or even weekly, me included.
- as I said in the begining, we already have matured, there is no need anymore for free client accounts. Maybe 50% discount for the first month or so. But no more free use.
- ask Synology or other NAS makers to offer a dedicated NAS model only for Storj, with all the apps and extrafeatures removed and blocked, at half the price. In my opinion, NASes are the most reliable machines for this type of service, that an usual small SNO can aquire, install and mantain.
Sorry if some of these ideas were told before, and sorry for the long post, but maybe you will find somethig usefull.
3 Likes
An increase in traffic volume is possible only with a proportional increase in HDD.
This is not possible on the same hardware.
It is possible if Storj manages to get some customers that actually use the stored data instead of using Storj just for backups (upload - wait - delete).
1 Like
I am in a different situation: while egrees is not metered, I’ve got a glass ceiling on egrees bandwidth. That is, nobody here will give me a better offer than what I pay for now. I’ve already experienced egrees peaks taking half of my egrees bandwidth. I’d be happy to host hundreds of TB, but at some point I’d have to shape the connection, that probably meaning losing most races, so essentially not getting paid for egrees at all.
I could go semi-pro and rent a data center slot. But that’s very costly, would essentially require some assurance from Storj that at least some risks of underutilization would be shared.
Each SNO may have work under different constraints. I’m therefore happy to see @john expressing interest in discussing cost structure.
I agree the payouts should not change.
But I hate to break this to you: STORJ is still in its infancy. At least in terms of market share.
STORJ’s stored customer data is less than 20 Petabytes. Backblaze, in comparison, has over 2 exabytes (2,000+ Petabytes). That is 100x larger and that’s just Backblaze. Google cloud, for example, can be another 10x.
Seeing the current rate of development, I too am bullish on the future. But as of now it hasn’t yet proportionatelly translated into growth in customers’ data and market share. Winning customers takes time.
With a very tiny piece of the pie, STORJ is what they call a “price getter”, rather than a “price setter”. It still has to play under the big guys’s dominance.
Again, I agree the payouts should not be reduced. Just wanna present the facts so everyone here aware of STORJ’s position and adjust the expectation of what it can do.
2 Likes
@John has mentioned another point that I would like to add to the discussion because it might give Storj some more good ideas.
He mentioned the question how to signal to SNOs that more capacity is needed and make them scale up (quickly). Imagine the situation where they sign a customer who requires a lot of space on the network.
Let’s put aside nodes that only have to change the settings value because they have more capacity already than they are reporting to the network. Let’s look at node operators who would have to invest to create new capacity, be it by investing time to build a new rig or to move data around or by investing money by purchasing larger disks.
I was asking myself what would be required so that I would go and buy let’s say a 20TB HDD for 500 Euros in advance and couldn’t give a good answer to that to myself. Nor to Storj.
So I think it would be interesting to hear what signals would others make them to invest upfront.
1 Like
For me it would be enough if Storj said something like “we are signing up a big customer who will be uploading a lot of data in a month, it will be about 10TB for each node and you’ll get a lot of egress from it”. As I have less empty space than that I would start looking for a new set of 6 drives to add to my server.
Normally I would expand the pool only if I was almost running out of space and it would take a bit of time to get the drives and test them (I got a bad batch of drives once so now I’m careful)
I have thought about something like that too. But is data so equally distributed that I could rely on such a statement? And how much egress can be expected (as we usually see a drop of egress when a drive is full). And for how long will such a customer stay with Storj?
What I am trying to say, such a statement would be helpful, but not at all guarantee that my hypothetical invest of 500 Euros ever will pay off.
I usualy add hdd when i have free used HDDs, as my work allow to have them time to time.
but they only 3-4 TB in size.
As a general approach to decreasing payments to SNO’s in order to make it economically viable for Storj it may be optimal to decrease payment in accordance to increasing demand for Storj from customers, meaning that net amount, paid to Storj node operators would be close or slightly above what we are seeing now. This could continue until income and costs reach some kind of balance. In such case though, small or recently started nodes would be in disadvantage.
Assuming the new customer uploads so much data that Storj needs nodes to expand before running out of space, I’d say the data would be equally distributed. Especially as smaller nodes fill up, larger nodes would get more than the 10TB or whatever.
Egress depends on the usage patters of the customer. For example, someone uses Storj for backups, then it’s upload → wait → delete and almost no egress.
Customers seem to be accessing new data more than old data, that’s why a full node gets less egress.
However, this hypothetical customer may access all of his data equally, access old data more than new data etc.
1 Like
I don’t know if it is available from everywhere, but on aliexpress you can get old xeon (12/24) + ~32-64Gb ECC REG ram + mobo with ~8 sata slots and pci-ex for further expansion for like 600-800$. There is also a pc case ( Fractal Design DEFINE R6) with 11 3,5" slots. It is quite a good deal compared to most NAS sotuions, imo.
This is why I suggested the auction contracts in post #4.
Have you considered selling them and getting larger drives instead, for lower power consumption/slot usage? Used drives sell well, at least in my locality.
With the correction that if your node is far away from the uploader, it will lose upload races more often.